VoiceBase AI takes a different approach to speech analytics, by leveraging patterns and data not recognizable to the human eye. We combined big voice data and machine learning technology, to automatically train predictive models from custom pre-tagged call data. This innovative solution detects complex events and predicts future behavior with extreme precision and accuracy.

Use finely tuned predictive models to
automatically score your calls

Icon of credit cards that shows how VoiceBase’s call center monitoring software redacts sensitive information

Payment Card Info

Hot Leads

Up-sell Probability

Social Security Numbers

Potential Churn

Caller Intent

Is this customer about to cancel their account?

Where artificial intelligence, becomes actionable.

Get Started
computer icon to show that you can integrate the rest api

Working With Predictive Analytics (Classifiers)

Using Classifiers
Listing Classifiers
Sample Results

Using Classifiers

VoiceBase runs the classifier while the recording is processed, returning a predicted class and confidence score. Common use cases of classifiers include:

  • Identifying leads and prospects
  • Automating quality control
  • Predicting customer churn
  • Detecting appointments, estimates, and orders
  • Marking calls sent to voicemail

Listing Classifiers

A list of classifiers available in your account can be obtained as follows by making a GET request on the /definitions/predictions/models resource. Here is a sample response.

Sample Results

Once a classifier is enabled on your media, there will be a prediction section included in the JSON response, here is an example.

Read the Docs >

AI

Q: How can I get started with Predictive Insights?

Q: How can I get started with Predictive Insights?

Reach out to VoiceBase to talk to a sales executive. Together we will define your needs and the predictive analytics models that will work to deliver the best results.

Q: What types of events can I detect?

Q: What types of events can I detect?

The events predictive models can predict is only limited to the events you have training data for. For example; if you want to predict how likely an upset customer is to cancel their account, VoiceBase will need 1,000 samples of the last call customers made before canceling, and 1,000 samples of other upset customers who’s accounts are still active.

Q: Can I use Speech Analytics instead of Predictive Insights?

Q: Can I use Speech Analytics instead of Predictive Insights?

Speech analytics is ideal to answer a question an analyst knows the answers to and can define, ex; Was an appointment made? What was the call driver? Was the customer upset?. AI Predictive Insights should be trained when it’s difficult for an analyst to define the answer to the question they are asking, ex; Why was this a hot lead? Is this customer going to churn? How likely is this caller to buy a 2nd item? Which item?

Q: What do I need to train a Predictive Insights Model?

Q: What do I need to train a Predictive Insights Model?

VoiceBase will need 1,000 pre-tagged calls for each potential result to the question you are trying to answer. For example, if the question is; Is this customer upset? Then there are two potential answers, yes or no. So we will need a minimum of 2,000 calls. If the question is; Why is this customer upset? And there are 10 different reasons your customers call and are upset, then VoiceBase will need a minimum of 10,000 calls, 1,000 for each potential result.

More FAQs >